Achilles and the Tortoise

Aristotle's own reply to the Achilles is one line long: it is the same as his reply to the Dichotomy. He says the argument assumes that it is impossible to traverse infinitely many things in a finite time, and that this is false, because the time is divided in just the same way as the distance. Aristotle adds that the Achilles differs from the Dichotomy only in that the successive magnitudes are not obtained by halving. He was right that the two arguments stand or fall together, and modern commentators agree.

That leaves three things genuinely worth arguing about.

Does the head start matter? In the Dichotomy the runner faces a fixed distance and subdivides it. In the Achilles there is a second body, and the subdivision is generated by the chase. Some commentators — Jonathan Barnes among them — hold that this makes the Achilles the more compelling of the two, because the stages are produced by the situation rather than imposed by the arguer. The mathematics is identical; the rhetoric is not.

Does summing the series actually answer it? This is the point on which the "calculus solved Zeno" story oversells itself. Summing the series shows that Achilles catches the tortoise at 11.111… seconds. It does not, by itself, explain how a process consisting of an endless sequence of stages with no final member gets completed. If you think that a run is nothing over and above being at various places at various times — Bertrand Russell's at-at account, developed by Wesley Salmon — then the question dissolves, because there is no sequence of achievements to complete. If you think a run is a doing, made of doings, the question remains open. Max Black argued in the early 1950s that the notion of completing an endless sequence of tasks is incoherent no matter what the distances sum to; Adolf Grünbaum (1967) argued at book length that it is not.

Was Zeno arguing for anything? The usual view, following Plato's Parmenides, is that Zeno was not defending the tortoise. He was defending his teacher Parmenides, who held that reality is one and unchanging, by showing that the ordinary pluralist picture generates absurdities too. On that reading the Achilles is not a puzzle awaiting a solution but a weapon in an argument, and its author would have been unmoved by being told that the series converges: the point was that your own assumptions produce this. Whether Zeno actually held that motion is impossible, or was only running a reductio, is disputed and the evidence does not settle it.

The evidence. Zeno of Elea wrote a single book, around 450 BCE, and it is lost. The Achilles survives in one place: Aristotle's Physics VI.9, at 239b14, where it is summarised in about forty words and immediately refuted. Aristotle introduces it as "the so-called Achilles", which tells us the argument already had a nickname a century after Zeno wrote — and that Aristotle expected his audience to recognise it.

Two things about that text are worth knowing. First, Aristotle describes only "the swiftest" and "the slowest". There is no tortoise in Aristotle. The animal appears in the later commentary tradition, and it is so firmly attached now that the paradox is unusable without it. Second, Aristotle's paraphrase is a hostile summary by someone with a theory to defend, not a quotation.

The handful of verbatim lines of Zeno that survive were quoted by Simplicius in his commentary on the Physics, written around 540 CE — roughly a thousand years after Zeno. All of them belong to Zeno's arguments against plurality. Not one surviving word of Zeno is about Achilles or the tortoise. Everything in this entry above the History section is reconstruction from Aristotle.

The tradition. Diogenes Laërtius (IX.25–29) supplies Zeno's biography: pupil and adopted son of Parmenides, an Eleatic, and a martyr in a plot against a tyrant who bit off his own tongue under torture. Diogenes gives incompatible versions of the last part, and it should be read as legend.

Plato's Parmenides (127a–128e) is the source for Zeno's purpose — a young man's defence of Parmenides against ridicule — and Plato is dramatising, not reporting.

Afterwards. The Achilles ran through the Greek commentators, through medieval discussions of the continuum, and back into the modern canon with Bayle's Dictionnaire in 1697. It picked up its reputation as the paradox everybody can state and nobody can quite dismiss. The mathematics that dispatches the series arrived with Cauchy in the 1820s and Weierstrass after him; the philosophical argument about whether the mathematics is the whole answer began again in the 1950s with Max Black and James Thomson and has not stopped.

It is the standard demonstration that an infinite process can have an end. Almost everyone meets the idea of a convergent series through some version of this race. The Achilles is the reason that the sentence "infinitely many things, finite total" is intuitive to anyone at all, and that sentence underwrites calculus, probability, and every numerical method in use.

It isolates a specific and common logical error. Proving something at every member of a sequence is not proving it at the sequence's limit. Written out that way it sounds like a technicality. It is the difference between a bank account that runs dry and one that does not, between an algorithm that terminates and one that stalls, and between Achilles winning the race and losing it. Undergraduates in analysis get this wrong every year, in exactly the form Zeno gave it.

It made "the slower is always ahead" into a testable claim. Zeno's real achievement was turning an obvious fact about the world into an argument with premises, so that someone could go and find which premise was wrong. It took about two thousand three hundred years. That is a long time, but the process only became possible because the argument was stated sharply enough to be attacked.

It keeps the honest version of the story available. The Achilles is the best available counterexample to the idea that mathematics finishes philosophy. The series converges; that is not in doubt and never was after 1830. Whether the convergence explains how an endless sequence of stages gets completed is still argued about by people who understand the mathematics perfectly well.

Convergence rates in computation. An iterative solver is Achilles and the answer is the tortoise: each pass closes some fraction of the remaining gap. Whether it arrives depends on the ratio, exactly as in the race — a solver with a contraction factor just under 1 converges but may take longer than the age of the universe to be useful, which is the practical version of the tortoise's taunt.

Pursuit problems in control and robotics. Interception, missile guidance and target-tracking algorithms compute precisely the quantity Zeno's argument avoids naming: the closing speed and the time to intercept. The naive "steer at where it is now" strategy is the tortoise's own description of the chase, and it works only because the gap shrinks geometrically.

Zeno points in simulation. Hybrid-systems simulators can hit states where events accumulate infinitely fast in finite simulated time — a bouncing ball whose bounces converge. Engineers call these Zeno points and design around them, because a simulator that reaches one makes no further progress while its clock stands still.

Financial and physical limits. A perpetuity paying a fixed fraction each period has a finite present value for exactly the reason Achilles wins: the terms form a convergent geometric series. The same computation prices an annuity, sums a multiplier effect in macroeconomics, and totals the light reflected between two facing mirrors. It is one formula and Zeno is the reason anyone thought it needed justifying.

The stage sequence

Let Achilles have speed $v_A$, the tortoise speed $v_T$ with $0 < v_T < v_A$, and let the head start be $d > 0$. Write $r = v_T / v_A \in (0,1)$.

Stage $0$ ends when Achilles reaches the tortoise's starting point:

$$t_0 = \frac{d}{v_A}$$

During $t_0$ the tortoise advances $v_T t_0 = d r$, which is the new gap. By induction the gap after stage $n$ is $d r^{,n+1}$ and

$$t_n = \frac{d}{v_A} r^{,n}$$

The total elapsed time over all stages is a geometric series:

$$T = \sum_{n=0}^{\infty} \frac{d}{v_A} r^{,n} = \frac{d}{v_A}\cdot\frac{1}{1-r} = \frac{d}{v_A - v_T}$$

which is exactly the closing-speed answer. Achilles' position at that moment is

$$x = v_A T = \frac{v_A, d}{v_A - v_T}$$

and the tortoise's is $d + v_T T$, which equals the same value. With $d = 100$, $v_A = 10$, $v_T = 1$: $T = 100/9 \approx 11.11$ s and $x = 1000/9 \approx 111.11$ m.

Where the inference fails

Define $s_N = \sum_{n=0}^{N} t_n$, the clock reading at the end of stage $N$. The argument establishes:

$$\forall N \in \mathbb{N}: \ x_A(s_N) < x_T(s_N)$$

and concludes:

$$\forall t \geq 0: \ x_A(t) < x_T(t)$$

The inference is invalid because ${s_N : N \in \mathbb{N}}$ is not all of $[0,\infty)$. It is a strictly increasing sequence bounded above by $T$, with $\sup_N s_N = T$ and $s_N < T$ for every $N$. The set of stage-endings has order type $\omega$ and its limit point $T$ is not a member of it. Every claim the argument proves is a claim about times strictly less than $T$; the overtake happens at $T$ and thereafter.

This is the cleanest diagnosis available, and it requires nothing about the metaphysics of motion — only the distinction between a property holding at every member of a sequence and holding at the sequence's limit. The same distinction is what makes pointwise convergence fail to preserve continuity, and it is one of the recurring sources of error in real analysis.

The residue

Suppose all of the above is granted. Two questions survive.

Order type. The stages form an $\omega$-sequence: there is a first, each has an immediate successor, and there is no last. If traversal really consists of performing each stage, then Achilles performs an $\omega$-sequence of acts and then is past them — a supertask. James Thomson (1954) argued that supertasks are incoherent, using a lamp switched at $t = 1 - 2^{-n}$ whose state at $t = 1$ is neither on nor off. Paul Benacerraf (1962) answered that the specification simply fails to determine the state at the limit, so nothing contradictory follows. Benacerraf's reply is generally accepted, and it establishes consistency rather than mechanism. The Achilles is safe either way, because unlike the lamp, Achilles' position is given by a continuous function whose value at $T$ is determined by the values before it: $\lim_{t \to T^-} x_A(t) = x_A(T)$. Continuity is doing the work that the lamp lacks, and saying so is a real explanation rather than a restatement.

Constant speed. The clean series above assumes uniform motion. Nothing in Zeno's argument requires it: the decomposition into stages works for any pair of motions in which the pursuer is faster. If both runners accelerate in the right coordinated way, the sum of stage durations can diverge and Achilles genuinely never catches up. That case is not a paradox — it is a slower pursuit than it looks — but it shows that "there are infinitely many stages" is compatible with both outcomes. The convergence, not the infinity, is what decides the race, and Zeno's argument never mentions the convergence at all.

A different tortoise

Lewis Carroll's "What the Tortoise Said to Achilles" (Mind, 1895) borrows the cast for an unrelated and equally durable problem. Carroll's tortoise accepts premises $A$ and $B$, and accepts that $A$ and $B$ together imply $Z$, but refuses to infer $Z$ until that implication is itself added as a premise $C$ — which generates $D$, and so on without end. The moral is that a rule of inference cannot be replaced by a premise, and any attempt to justify inference by adding premises regresses forever. Douglas Hofstadter revived both tortoises in Gödel, Escher, Bach (1979), which is where most readers now meet either. The two paradoxes have nothing in common but the animals.